Bathymetry and water column correction
WorldView 2 image Princess Cays, Eleuthera island, Bahamas

  November 11th 2011, 2 m ground resolution
image from Digital Globe
"(c) DigitalGlobe, Inc. All Rights Reserved"
"Includes copyrighted material of DigitalGlobe, Inc., All Rights Reserved"
work done in 2012, 2013 and 2014

The "PAN solution"
uses the PANCHRO bans, 4*4 sum-aggregated and co-registered

bands 1, 2, 3, Panchro, 4, 5, 6, 7 and 8       become      bands 1, 2, 3, 4, 5, 6, 7, 8 and 9


August 2014
Optical calibration of the PAN band

  • Now that the PAN solution using WV2 imagery is established, it still has to be illustrated and made into a practical operational process
  • The key element is the precise description of the variation of the operational diffuse attenuation coefficient for the WV2 Panchromatic band as water depth increases in this image
    • this is a dimensionless relationship, which is based on optical path in the Blue and PAN bands
  • I now have designed in 4SM code an automated way to do that
    • it requires a carefull preparation of the special effect mask
    • using this WV2 image of Princess Cays, this in the end yielded a fairly reasonable result, as illustrated below
      • this plot mainly shows the curved "exponential decrease" of the linearized Xpan=log(Lspan-Lswpan) and Xblue=log(lsblue-Lswblue), as water depth increases over the brightest bottom type
      • this plot also shows all isobath lines at 1 m interval, from 0 to 30 m
    • this result still needs to be refined using a few more WV2 images over scenes where clean bright sands are more prevailing: it should be a very smooth curve, without any "road humps"

The Blue vs PAN observed optical model
As of september 2014,
the Blue vs PAN optical model
is now computed, using
  • WV02 response curve for the Panchro band, 
  • Jerlov's data for the observed water type

The Blue vs PAN computed optical model
september 2014


WZ=all solutions WZ=PAN solution only

image WZ
Black....solution by the RedEdge band
Red.......solution by the Red band
Yellow...solution by the Yellow band
Grey......solution by the PAN band

image WZ
Grey.....solution by the PAN band
normalized TCC water column corrected
using all solutions
  • some areas exhibit weird results, likely caused by locally turbid waters: these areas yield underestimated depth results (and therefore darker water column corrected radiances)

normalized TCC water column corrected
using only the PAN solution
  • some areas exhibit weird results, likely caused by locally turbid waters
    • this is a remarkable feature of the PAN soliution in 4SM: areas of questionable results are highlighted in dark purple in this normalized water column corrected display
  • the PAN solution yields more homogeneous water column corrected results
    • this improves the potential for bottom typing

image Z250
using all solutions

image Z250

using only the PAN solution

Benefits of the PAN solution
  • Using the PAN solution is feasible and profitable: only uses Coastal, Blue, Green and PAN bands for retrieving depth
  • Based on the optical calibration for the MULTI bands, the optical calibration of the PAN band is derived, using the specific PAN response curve and Jerlov's data.
    • this is a now generic and automatic process, hard coded in 4SM, which requires minimal manual control
    • it may be used for any brand of imageries, only requires the specific PAN response curve
  • It yields very detailed results in the shallow range 0-5 m 
  • It yields reasonably smooth results over the deeper range, thanks to the following:
    • the level of system noise is quite low in this image, apart from vertical striping
    • the computed optical model for the PAN band is a continuous smooth curve
  • No more of these annoying gaps that show up when using the MULTI solutions: removal of the water column effect is much improved

BOA normalized TCC water column corrected (blue, green and red)
using only the PAN solution
The footprint of UKHO's MBES coverage is shown

BOA normalized TCC water column corrected (blue, green and red)
over the MBES area

Retrieved depth

Note that
  • The result is remarkably smooth,  thanks to the Blue vs PAN computed optical model for PANchro band: no more "road humps"
  • Water column correction is complete, as the retrieved depth image does not exhibit any trace of the bottom type variations

I wish I had access to UKHO'S MBES dataset, property of Princess Cruise Lines

Download from my GoogleDrive (123 MB)
  • princesscayswvmp_sept2014_WCC.tif....................BOA watercolumn corrected radiances, from 1 to 9: channel4 is PAN
  • princesscayswvmp_sept2014_Zcm.tif......................Retrieved depth in centimeters
  • princesscayswvmp_sept2014_BottomType.tif..........Bottom type classification (step 2 final)

Command Line september 2014
nice -20  ./4SM  -Process  -Origin/DigitalGlobe                       
-Model/v/runWV2/makeClouds/mask_3  -Tpsv/2/4/Text/MSE_3/Bmin100/panCode1.08   
-Smooth/5/Csmooth4/D/Smart+    @-Eshp    -LL10    @-DNshp-DDshp        
@-DD1686/1743    @-E1/6001/1/200 @-E1/6001/900/5701    @-DN4456/493    
@ProfileAB/LzrvWb/profile_white/dTM/chAB_6_7 _2_3_4           
@-ProfileAB/lZrvWB/profile_purple/DTM/chAB_1_2_3 _5_6           

Créer un site
Créer un site